34028 research outputs found
Sort by
Quinolone as a Privileged Scaffold: A Brief Overview on Early Classical and Recent Advanced Synthetic Pathways, Innovative Neuroprotective Potential, and Structure‐Activity Relationships
No full text Neurodegenerative disorders represent the second largest group of diseases worldwide. 2(1H)-quinolone and its structural congener, 4(1H)-quinolone have recently become significant topics in the field of drug design and development of modulators of neurotransmitter systems and neuroprotective agents to tackle neurodegenerative disorders. In this review, the structural properties and the early classical as well as the recent novel synthetic strategies for 2(1H)-/4(1H)-quinolone are discussed. The neuropharmacological activity and mechanisms of action of several 2(1H)-/4(1H)-quinolinone-based compounds are demonstrated with special emphasis on the structure–activity relationships (SAR). Therefore, the perspectives elaborated in this review could guide medicinal chemists for rational design and development of novel 2(1H)-/4(1H)-quinolone therapeutic candidates targeting neurodegenerative diseases.</p
Digital Praise in Professional Networks AI-Generated LinkedIn Posts as Contemporary Expressions of Epideictic Rhetoric GLOWKA
No full textThe study examines AI-generated LinkedIn posts through the lens of epideictic rhetoric, analysing how these digital artifacts reshape professional self-presentation and redefine notions of professional excellence. It explores how AI algorithms create systematic suggestion patterns that influence norms of praise and value expression in professional discourse. The study offers insights into how artificial intelligence is transforming rhetorical practices in professional settings and raises questions about authenticity and the algorithmic mediation of human experience.</p
Weakly supervised lesion segmentation via principal axis estimation
No full textState-of-the-art lesion segmentation methods often rely on large quantities of manu?ally annotated data to produce acceptable results. In settings where such labelled data may be scarce, there may be value in exploiting data that is cheaper to ac?quire or more readily available through clinical trials, such as the bilinear principal axes measurements Response Assessment in Neuro-Oncology (RANO) and Response Evaluation Criteria in Solid Tumours (RECIST). This work demonstrates the utility of such measurements for medical image segmentation, whereby an encoder network is first trained to regress principal axis measurements, using Mean Squared Error (MSE) and cosine similarity loss, which promotes orthogonality of the principal axes. Additionally, using Axis-Aligned Bounding-Boxes (AABB) to measure overlap with the Ground Truth (GT), the encoder model is shown to estimate tumour principle axes with good performance reliably. The trained encoder was combined with a Randomly Initialized (RI) decoder for fine-tuning as a U-Net architecture for lesion segmentation. The results demonstrate that models trained with a weakly super?vised encoder converge faster than those without pre-training, converging within 7 epochs using a Pre-trained Trainable Encoder (PTE) model compared to 13 epochs using a Randomly Initialized (RI) model. These results alongside successful training on only 10% of data show that pre-training helps to minimise the annotation bur?den when performing segmentation. The wider implications of this work encompass the ability to leverage previously unknown and underutilized resources. This allows an improvement in application quality for situations that rely on low amounts of high-quality data annotation, particularly in the case of medicine.</p
Emotional labour in teaching
No full textOverview - Teacher recruitment and retention in England has been problematic since 2011, despite the recent pay increases. Emotional labour is a well-researched concept in the service sector and has been given some attention in academic research in education. Yet, teachers’ understanding of it is under-researched. This study has discovered that teachers have little theoretical understanding of emotional labour and even less training in how to manage this aspect of their job. Since teachers consider emotional labour to be mainly centred around the care and emotional connection they have to their work and their students, they risk burn-out and compassion stress injury, both of which can hinder wellbeing.Methodology - interview and focus groups.Analysis - thematic analysis</p
Acceptability of Remote Monitoring Technologies for Early Warning of Major Depression
No full textRecent research shows that 16% of individuals aged 16 and older are affected by depression, with Major Depressive Disorder (MDD) severely impairing daily life and potentially leading to suicide. Early intervention is vital, but current early warning methods of depression are time-consuming and not scalable. This study examines the acceptability of smartphone-based early warning systems using machine learning for depression intervention. Interviews were conducted with participants from the RADAR-MDD study, where smartphone sensors collected behavioural data. Three key themes emerged: designing remote monitoring technologies (RMTs) to enhance user engagement and wellbeing, RMTs as tools for comprehensive and empowering mental health support, and ethical and purpose-driven data utilization by RMTs. Whilst participants were open to RMTs, they expressed concerns about data commercialization. The study highlights the importance of prioritizing user experience, ethics, and personalization in designing effective early warning systems. </p
ALMAGAL: III. Compact source catalog: Fragmentation statistics and physical evolution of the core population
No full textThe physical mechanisms behind the fragmentation of high-mass dense clumps into compact star-forming cores and the properties of these cores are fundamental topics that are heavily investigated in current astrophysical research. The ALMAGAL survey provides the opportunity to study this process at an unprecedented level of detail and statistical significance, featuring high-angular resolution 1.38 mm ALMA observations of 1013 massive dense clumps at various Galactic locations. These clumps cover a wide range of distances (~2–8 kpc), masses (~102–104 M⊙), surface densities (0.1–10 g cm‑2), and evolutionary stages (luminosity over mass ratio indicator of ~0.05 < L/M < 450L⊙/M⊙). Here, we present the catalog of compact sources obtained with the CuTEx algorithm from continuum images of the full ALMAGAL clump sample combining ACA-7 m and 12 m ALMA arrays, reaching a uniform high median spatial resolution of ~1400 au (down to ~800 au). We characterize and discuss the revealed fragmentation properties and the photometric and estimated physical parameters of the core population. The ALMAGAL compact source catalog includes 6348 cores detected in 844 clumps (83% of the total), with a number of cores per clump between 1 and 49 (median of 5). The estimated core diameters are mostly within ~800–3000 au (median of 1700 au). We assigned core temperatures based on the L/M of the hosting clump, and obtained core masses from 0.002 to 345 M⊙ (complete above 0.23 M⊙), exhibiting a good correlation with the core radii (M ∝ R2.6). We evaluated the variation in the core mass function (CMF) with evolution as traced by the clump L/M, finding a clear, robust shift and change in slope among CMFs within subsamples at different stages. This finding suggests that the CMF shape is not constant throughout the star formation process, but rather it builds (and flattens) with evolution, with higher core masses reached at later stages. We found that all cores within a clump grow in mass on average with evolution, while a population of possibly newly formed lower-mass cores is present throughout. The number of cores increases with the core masses, at least until the most massive core reaches ~10M⊙. More generally, our results favor a clump-fed scenario for high-mass star formation, in which cores form as low-mass seeds, and then gain mass while further fragmentation occurs in the clump.</p
Implementation of NGS and SNP microarrays in routine forensic practice: opportunities and barriers
No full textForensic DNA analysis plays a pivotal role in personal identification, kinship assessment, and criminal investigations, with Short Tandem Repeat (STR) typing via capillary electrophoresis (CE) long established as the gold standard. However, CE-based STR analysis faces notable limitations in multiplexing capacity, the interpretation of degraded or mixed samples, and the resolution of complex kinship relationships. Emerging technologies such as Next Generation Sequencing (NGS) and Single Nucleotide Polymorphism (SNP) microarrays present promising alternatives that can address these shortcomings and expand the scope of forensic DNA testing. Despite their potential, the adoption of these methods in routine forensic practice remains limited due to high costs, technical complexity, and a lack of standardised protocols and legal frameworks. This review critically examines the capabilities, limitations, and current applications of NGS and SNP microarrays in comparison to traditional STR CE profiling. NGS enables STR sequencing and SNP typing with enhanced discriminatory power, better performance with degraded DNA, and improved mixture deconvolution. Conversely, SNP microarrays offer a cost-effective solution for extended kinship testing, Forensic Investigative Genetic Genealogy (FIGG), and phenotypic prediction, though they are less effective with low-quality samples and DNA mixtures. Ethical, legal, and privacy concerns, particularly surrounding the use of Forensic DNA Phenotyping (FDP) and consumer genetic data in FIGG, further complicate their integration into forensic workflows. While significant challenges remain, technological advancements and growing regulatory efforts point towards an achievable path for wider implementation. A hybrid approach that combines STR CE for routine casework with NGS and SNP microarrays for complex scenarios, supported by investments in bioinformatics training, database expansion, and ethical governance, offers a practical strategy for integrating these technologies into future forensic practice. </p
Towards Scientific Knowledge Graphs: Dependency Graph Analysis Using Graph Neural Networks for Extracting Scientific Relations
No full text
Scientific relation extraction plays a crucial role in constructing scientific knowl- edge graphs that can contextually integrate knowledge from the scientific literature. How- ever, a large majority of existing efforts do not support human guidance, which hinders refining the construction of scientific knowledge graphs and, thus, the natural cycle of scientific knowledge integration. Therefore, there is a necessity to ground the human– machine collaboration in learned mechanisms, the prerequisite of which is quantifying the contribution of candidate mechanisms. In addressing this, we introduce an efficient summation node architecture by leveraging a graph neural network (GNN) on semantic patterns among dependency graphs. Then, we quantify the potential of different semantic invariance in serving as semantic interfaces towards the flexible construction of scientific knowledge graphs. Specifically, we posit that collocation-level patterns can enhance both extraction accuracy and F1 scores. Our proposed solutions exhibit promising performances for certain relations under bi-classification configurations, facilitating the learning of more semantic invariance from the word level to the collocation level. In conclusion, we assert that the flexible and robust construction of scientific knowledge graphs in the future will necessitate continual improvements to augment learned semantic invariance. This can be achieved through the development of more integrated and extended input graphs and transformer-based GNN architectures. </p
Deep histological synthesis from mass spectrometry imaging for multimodal registration
No full textThis work proposes a solution that synthesises histological images from MSI, using a pix2pix model, to effectively enable unimodal registration.</p
Zero-Shot Sim-to-Real Reinforcement Learning for Fruit Harvesting
No full textThis paper presents a comprehensive sim-to-real pipeline for autonomous strawberry picking from dense clusters using a Franka Panda robot. Our approach leverages a custom Mujoco simulation environment that integrates domain randomization techniques. In this environment, a deep reinforcement learning agent is trained using the dormant ratio minimization algorithm. The proposed pipeline bridges low-level control with high-level perception and decision making, demonstrating promising performance in both simulation and in a real laboratory environment, laying the groundwork for successful transfer to real-world autonomous fruit harvesting.</p